How much do returns cost retailers?

US retailers absorbed $849.9 billion in returned merchandise in 2025, equal to 15.8% of all retail sales, according to the NRF. Beyond the lost sale, recovering value from returned goods costs an estimated $200 billion annually in labor, logistics, inspection, and liquidation, yet retailers recover only about 50% of a product's original worth through resale or refurbishment (McKinsey). Return fraud compounds the damage further, with Appriss Retail and Deloitte valuing fraudulent returns at $103 billion in 2024 alone. The true cost of returns is therefore closer to twice the sticker price of the merchandise that comes back.

What is the average return rate for online retailers?

The average online return rate in the US is 19.3%, meaning nearly 1 in 5 items shipped by e-commerce retailers comes back, compared to an overall retail return rate of 15.8% across all channels (NRF, 2025). Online return rates run higher than in-store rates because shoppers cannot physically inspect or try on products before purchase. Gen Z shoppers averaged nearly 8 online returns per person last year, and bracketing, the practice of buying multiple sizes or colors with the intention of returning most, is rising among nearly three-quarters of online sellers (McKinsey/NRF). The gap between online and in-store return rates is a direct measure of how much product information, fit tools, and address accuracy still matter.

What causes the most product returns in retail?

Size and fit is the single largest cause of returns in fashion retail, responsible for roughly 70% of the category's returns (McKinsey). The second major cause is products that do not match their page descriptions, including wrong dimensions, misleading imagery, or missing attributes, commonly called 'not as described' returns. Bracketing, where 63% of US shoppers admit to buying multiples planning to return some, is the third behavioral driver (ReturnPrime). Incorrect delivery addresses cause up to 41% of first-attempt delivery failures, which convert into returns or undeliverable shipments (Digital Commerce 360). Quality defects and fraud account for the remaining volume. Addressing these five root causes requires fixes upstream in the commerce stack, not at the returns desk.

How can retailers reduce e-commerce returns?

Retailers reduce e-commerce returns most effectively by attacking the root causes before the package ships. Accurate, AI-enriched product content eliminates 'not as described' returns; GSPANN's ContentHubGPT uses generative AI to produce SEO-ready product descriptions that match what shoppers actually receive, cutting manual content effort by more than 70%. AI-powered sizing tools and virtual try-on reduce fit-related returns by up to 40%, according to True Fit. Standardizing shipping addresses with transformer-based address normalization, a capability GSPANN deploys for omnichannel retailers, prevents returns caused by failed first-attempt delivery. AI fraud scoring flags suspicious return requests in real time, and computer vision in the reverse logistics warehouse routes returned items to their highest-value disposition path automatically.

What is dispositioning in reverse logistics, and why does it matter?

Dispositioning is the process of evaluating each returned item and routing it to its highest-value recovery path, whether that is restocking, refurbishing, liquidating, recycling, or breaking it down for parts. It is the most expensive and operationally complex stage of reverse logistics. More than half of supply chain executives name dispositioning as their top returns challenge, and US retailers spend approximately $200 billion annually on this process while recovering only about 50% of original product value (McKinsey). Computer vision systems now automate condition grading and routing decisions in real time, replacing slow manual inspection. Platforms such as Two Boxes process nearly $1 billion in returned inventory annually using this approach, turning the returns warehouse into an AI-powered sorting center.

Retail Returns: Five AI Fixes for a $850B Problem

Returns are retail's quiet profit killer. Virtual try-on gets the headlines, but it only fixes one reason people send things back.

Here is what really drives returns, what actually reduces them, and what the early adopters already know.

$849.9 Billion in Returns, and the Online Problem is Getting Worse

$849.9 billion in merchandise came back in 2025, or 15.8% of all US retail sales (NRF)
Online return rate is 19.3%, nearly 1 in 5 items shipped comes back
Gen Z shoppers averaged nearly 8 online returns per person last year (NRF)
9% of all returns are fraudulent (NRF); Appriss Retail and Deloitte put the 2024 dollar impact at $103 billion

Returns are not slowing down. They are accelerating, especially online, and fraud is compounding the damage.^{[1] & [2]}

hree retail returns statistics: $849.9B merchandise returned in 2025, 19.3% online return rate, $103B in fraudulent returns

Returns Trigger a Second $200 Billion Problem, and Retailers Recover Only Half the Value

US retailers spend an estimated $200 billion annually recovering value from returns, yet recover only about 50% of a product's worth via resale, liquidation, or parts (McKinsey)

Over half of supply chain executives call dispositioning their top returns challenge (McKinsey)

1 in 4 transactions includes a "bracketed" item (consumer behavior of purchasing multiple versions of the same product with the intention of keeping only one and returning the rest.), and nearly three-quarters of sellers say this is rising (McKinsey/NRF)^[3]

wo stat callouts: $200B annual cost to recover value from returns, approximately 50% of product value recovered via resale

Why Returns Actually Happen: Five Root Causes

Size and fit: largest cause of fashion returns, roughly 70% of the category (McKinsey)

“Not as described”: products that do not match photos and copy (ReturnPrime)

Quality and defects: items arriving damaged or inconsistent

Bracketing: 63% of US shoppers admit to buying multiples planning to return some (ReturnPrime); 45% say “bending the truth” is acceptable (NRF)

Incorrect delivery address: up to 41% of first-attempt delivery failures trace to address errors (Digital Commerce 360). Missed packages become returns, RTO, or escalations.^{[4], [5], [6] & [7]}

Bar chart: size and fit causes 70% of fashion returns, 63% of shoppers bracket, 41% of deliveries fail due to address errors

Product Data and Content Enrichment Attacks “Not as Described” Returns

What the product page says is what the shopper gets:

“Not as described” returns happen when the product page doesn't match what arrives — wrong dimensions, missing attributes, or imagery that misleads shoppers
AI-generated product content cuts manual content effort by 70%
ContentHubGPT uses generative AI and real-time trending keywords to turn raw product data into SEO and GEO ready descriptions in a fraction of the time, keeping every product page accurate, on-brand, and discoverable in both search results and AI shopping assistants
Accurate product data is the first line of defense against mismatch returns.^{[8], [9] & [10]}

AI Sizing and Virtual Try-on Attack Fit Returns Helping Shoppers See the Right Fit Before They Buy

Size and fit is the single largest cause of fashion returns, roughly 70% of the category (McKinsey)
Virtual try-on uses generative AI and body modeling so shoppers see fit before they buy; AI sizing engines recommend the right size from past purchases, garment data, and fit reviews
True Fit reports return-rate reductions of up to 40% from AI-powered fit tools

Stat callout: 70% of fashion returns are caused by size and fit; AI-powered fit tools cut return rates by up to 40%

AI-Driven Address Normalization Attacks Delivery Returns

Fixing the hidden driver of Returns to Origin before parcels leave the warehouse:

Up to 41% of first-attempt delivery failures trace to address errors, and missed packages become returns, Returns to Origin (RTO), or customer service escalations (Digital Commerce 360)
T5-based transformer models standardize fragmented address inputs into postal-authority format before the package ships
GSPANN has an AI and Machine Learning-based address normalization blueprint and deploys this for omnichannel retailers
Clean the address before the parcel moves. It is the cheapest return to prevent.^{[15] & [16]}

AI Fraud Detection Attacks the $103 Billion Fraud Leak

Stopping refund abuse without breaking the customer experience:

9% of all returns are fraudulent, and Appriss Retail and Deloitte put the 2024 dollar impact at $103 billion
AI models score every return request in real time using customer history, transaction patterns, and product signals to flag suspicious activity before a refund is issued
85% of retailers now deploy AI for return fraud scoring (NRF), enabling tiered policies where trusted customers keep frictionless returns and flagged customers face stricter terms

The goal is not to punish legitimate returns. It is to make fraud unprofitable while keeping the experience seamless.^{[17], [18] & [19]}

hree fraud stats: 9% of returns are fraudulent, $103B dollar impact in 2024, 85% of retailers use AI for fraud scoring

Computer Vision in Reverse Logistics Attacks Dispositioning Cost

Routing returned items to their highest-value path automatically:

Over half of supply chain executives call dispositioning their top returns challenge (McKinsey)
Computer vision auto-grades each returned item's condition and routes it to the highest-value path, restock, refurbish, liquidate, recycle, or parts, instead of defaulting everything to slow manual inspection
Two Boxes processes almost $1 billion in returned inventory annually (FreightWaves)

The returns warehouse is becoming an AI-powered sorting center.^{[21] & [22]}

Two stats: 50%+ of executives cite dispositioning as their top challenge; Two Boxes processes ~$1B annually

GSPANN'S PERSPECTIVE

Four Fixes to the Commerce Stack That Solve Returns at the Root

Poor product data and poor address data are the same problem: incomplete information, wrong outcome. Shoppers cannot picture the fit. AI agents cannot recommend. Parcels miss the door. Here is what we recommend for leaders rebuilding the stack underneath:

1. Make Every Product Page Tell the Truth: Descriptions, specs, and imagery that match reality across every channel mean shoppers buy with confidence and mismatch returns drop.

2. Let Shoppers See the Fit Before Checkout: Virtual try-on, AR, and AI sizing move “will this fit me” out of the return window and into the decision moment.

3. Clean the Address Before the Parcel Moves: Standardize every shipping address at capture, so packages arrive first time and Returns to Origin stop quietly eating margin.

4. Architect for Change, Not for a Single Vendor: A composable, API-first stack lets you add, swap, and upgrade layers as the market shifts, without replacing the core.

The retailers winning the next five years will be the ones whose product pages, try-on, fraud signals, and address data all speak the same language. That is how returns stop being a leaking cost and start becoming a competitive advantage.^{[23] & [24]}